Coronary microvascular ischemia in hypertrophic cardiomyopathy - a pixel-wise quantitative cardiovascular magnetic resonance perfusion study

Thirty-six patients with HCM referred for CMR at the Royal Brompton Hospital were studied. HCM was diagnosed in accordance with standard clinical guidelines [25]. Patients were excluded if they had: conditions associated with coronary microvascular dysfunction such as diabetes; significant epicardial coronary artery disease on angiography (defined as >50% diameter stenosis in a major coronary artery); previous gradient reduction therapy; contraindications to CMR, adenosine, or gadolinium-based contrast agents. The study was approved by the National Research Ethics Service and was conducted in accordance with the principles set out in the declaration of Helsinki, with written informed consent obtained from all patients.

All patients were asked to abstain from caffeine-containing beverages or drugs for 24 hours prior to imaging and from β-blockers and rate-limiting calcium channel antagonists for 48 hours prior to imaging. Images were acquired using a dedicated 1.5 T scanner with a twelve-channel phased-array receiver coil (Siemens Magnetom Avanto, Siemens AG Healthcare Sector, Erlangen, Germany). A retrospectively-gated balanced steady-state free-precession sequence was used to obtain breath-hold cine images in three long-axis planes, followed by a contiguous stack of short axis slices from the atrioventricular ring to the apex [26]. The end-systolic frames of the long-axis cine images were used to plan the acquisition of three short axis perfusion images to cover the base, mid-LV and the apex. Adenosine was infused at 140 mcg/kg/min for a minimum of 4 minutes to achieve hyperemia. After measurement of heart rate and blood pressure at peak stress, gadolinium contrast (Gadovist, Bayer-Schering, Berlin, Germany, 0.1 mmol/kg) was rapidly injected at 3.5 ml/s, followed by 15 ml saline at 7 ml/s using a power injector (Medrad UK, Ely, Cambridgeshire, UK) to ensure a compact bolus entered the heart. A saturation-recovery prepared dual-sequence approach with center-out hybrid echoplanar imaging (EPI) [27] was used for perfusion imaging with the following typical sequence parameters: fat saturation pulse, composite 90° saturation preparation pulse for each slice [28], 28° readout pulse, saturation recovery time to central raw data acquisition 90 ms, repetition time 5.1 ms, echo time 1.1 ms, echo train length 4, field of view 360 × 288 mm, base resolution 160 × 160, slice thickness 8 mm. The center frequency of the scanner electronics was manually tuned to water in a 10 × 10 × 10 cm volume encompassing the LV to optimise both fat suppression and center-out hybrid-EPI image quality [29]. The arterial input function slice used low-resolution fast low-angle shot (FLASH) imaging with an adiabatic B1-insensitive rotation type 4 saturation pulse [30].

Three short axis images and an image at the basal slice of the arterial input function were acquired every cardiac cycle for a minimum of 50 cycles. Two initial proton density-weighted images were acquired prior to the arrival of contrast as part of perfusion imaging and were used for subsequent surface coil intensity correction. After ~10 min, late gadolinium enhancement (LGE) images were acquired with an inversion recovery-prepared segmented turbo FLASH sequence [31]. Inversion times were optimised to null normal myocardium with images repeated in two orthogonal phase-encoding directions to exclude artifact. After a minimum of 30 minutes, rest perfusion imaging was carried out at the same slice positions.

Ventricular volumes, function, mass, and ejection fraction were measured using a semi-automated threshold-based technique (CMRtools, Cardiovascular Imaging Solutions, London). All volume and mass measurements were indexed to body surface area [32]. End-diastolic LV wall thickness was determined for each of the 17 American Heart Association (AHA) segments excluding the apex [33]. Late enhancement was dichotomously assessed for each segment by an expert reader blinded to the perfusion data and considered to be present if there was an area of high signal intensity on a background of adequately nulled myocardium present in two orthogonal phase-encoding directions [12].

Absolute MBF was quantified pixel-wise at rest and at peak stress as outlined in Figure 1 and as previously described [23]. In brief, endocardial and epicardial borders of the LV myocardium were manually traced using Argus CMR software (Syngo, Siemens Healthcare, Erlangen, Germany) to define myocardial regions-of-interest (ROI). Custom image processing software developed in the Interactive Data Language (Exelis Visual Information Solutions, Boulder, Colorado, USA) was used to correct surface coil-intensity bias and motion artifacts for each image series to ensure frame-to-frame correspondence of pixels. MBF was then quantified pixel-wise using model-constrained deconvolution as previously validated [23].

To avoid potential underestimation of the severity of perfusion defects in a sector-wise analysis, ROI analysis was performed using the MBF pixel maps to compare hypoperfused areas and remote hyperemic areas at stress to corresponding areas at rest. Hypoperfused areas were defined as areas which visually appeared to have the worst perfusion on stress perfusion pixel maps. Based on the minimum myocardial perfusion reserve index (MPRI = stress MBF/rest MBF) as measured from these ROI, the cohort was divided into two groups for further comparison: severe microvascular dysfunction (defined as minimum MPRI < 1.0) and non-severe groups (minimum MPRI ≥ 1.0).

To assess the relationship between perfusion, wall thickness, and the presence of late enhancement, the myocardium was also divided into 16 segments according to the 17-segment AHA model, omitting the apex. Segments were further divided into endocardial and epicardial layers to assess for transmural perfusion gradients [33]. Perfusion in areas of LGE was also compared with that in remote areas free of enhancement.

Continuous variables are expressed as mean ± standard deviation (SD) for normally distributed variables and as medians with interquartile ranges for non-parametric data. The Kolmogorov-Smirnov test was used together with histograms to assess the normality of continuous data. Differences between parametric continuous variables were assessed using Student’s t-test, and for non-parametric data, the Mann–Whitney U-test. Categorical data are presented as frequencies and percentages. Differences between categorical variables were assessed using the χ² and Fisher’s exact tests as appropriate.

To take into account correlation of repeated measurements and clustering of data from ROI and sectors within slices and patients, the perfusion data was analysed using a multilevel linear mixed effects model with patients treated as a random intercept. The relationship between the presence of LGE and perfusion was assessed using binary logistic regression together with a mixed effects multilevel generalised linear model. Two-tailed values of P < 0.05 were considered significant. Statistical analysis was performed using Stata SE Version 12 (StataCorp, College Station, Texas, USA).

The final study cohort was comprised of 35 patients (Table 1). One patient was excluded due to an inadequate vasodilator response to adenosine, which on further questioning was due to caffeine ingestion, leaving a final cohort of 35 patients. In response to adenosine, resting heart rate rose from a mean of 69.7 ± 10.3 to 92.6 ± 13.6 beats per minute with a small fall in mean arterial pressure from 94.8 ± 11.7 to 90.3 ± 11.0 mmHg. Six out of 560 segments were excluded from analysis due to imaging artifact, encroachment of the left ventricular outflow tract on the basal slice, or problems with surface coil intensity normalisation. Overall, 99 segments had LGE.

Examples of perfusion pixel maps for the severe microvascular dysfunction and the non-severe patients are presented in Figure 2. Based on ROI analysis of the most significant perfusion defects seen in each patient, 11 (31.4%) patients showed evidence of severe microvascular dysfunction. There were no significant differences between the severe and non-severe groups with respect to baseline clinical and demographic features, although there was a trend towards a higher incidence of left ventricular outflow tract obstruction amongst those with severe microvascular dysfunction (Table 1). However, with respect to the CMR findings, the maximum end-diastolic wall thickness was significantly higher in the severe microvascular dysfunction patients versus the non-severe patients (Table 2).

ROI analysis of hypoperfused areas in the severe microvascular dysfunction patients (Figure 3A) revealed that mean resting MBF was significantly lower in hypoperfused areas relative to remote areas from the same slices (1.22 ± 0.36 ml/g/min versus 1.39 ± 0.34 ml/g/min, P < 0.001). After stress, whereas MBF rose significantly in the remote areas (1.39 ± 0.34 ml/g/min rising to 2.60 ± 0.57 ml/g/min, P < 0.001), stress MBF in hypoperfused ROI in patients with severe microvascular dysfunction not only failed to rise, but fell significantly from baseline values (1.22 ± 0.36 ml/g/min falling to 1.05 ± 0.39 ml/g/min, P = 0.021). In contrast, for the non-severe patients (Figure 3B), there was a significant rise in mean MBF with stress, even in the hypoperfused areas (1.05 ± 0.28 ml/g/min rising to 1.87 ± 0.45 ml/g/min, P < 0.001). ROI analysis in both groups showed significant hyperemic responses in regions remote from perfusion defects (1.39 ± 0.34 ml/g/min rising to 2.60 ± 0.57 ml/g/min, P < 0.001; and 1.20 ± 0.31 ml/g/min rising to 2.74 ± 0.85 ml/g/min, P < 0.001, respectively). The ratio of the minimum MPRI for hypoperfused areas to the maximum MPRI in remote regions was significantly lower in the severe microvascular dysfunction patients at 0.31 ± 0.10 versus 0.58 ± 0.16 in the non-severe group (P < 0.001).

Adenosine achieved hyperemia, with MBF rising significantly between rest and stress. When considering the most normal sectors in each of the patients, the mean MPRI was significantly lower in the severe versus non-severe patients for sectors as a whole and in endocardial sub-sectors, with a trend towards significance in the epicardium (Figure 4). When comparing the 16 segments in all the patients, MBF rose significantly with stress in whole sectors (1.22 ± 0.34 ml/g/min rising to 2.22 ± 0.76 ml/g/min, P < 0.001) and in both endocardial and epicardial subsectors (1.25 ± 0.35 ml/g/min rising to 2.00 ± 0.76 ml/g/min, P < 0.001; and 1.20 ± 0.35 ml/g/min rising to 2.36 ± 0.83 ml/g/min, P < 0.001, respectively). At rest, endocardial MBF was significantly higher than epicardial MBF (1.25 ± 0.35 ml/100 g/min versus 1.20 ± 0.35 ml/g/min, P < 0.001) with an endocardial to epicardial MBF ratio of 1.05 ± 0.11. However, at stress, this pattern was reversed with endocardial MBF increasing significantly less than epicardial MBF (2.00 ± 0.76 ml/g/min versus 2.36 ± 0.83 ml/g/min, P < 0.001) giving a ratio of 0.85 ± 0.18. The more blunted hyperemic response in the endocardium relative to the epicardium was also reflected by a significantly lower mean MPRI (1.68 ± 0.65 versus 2.06 ± 0.73, P < 0.001 respectively).

There was no significant relationship between resting MBF and sector end-diastolic wall thickness. However, at stress, there was a negative correlation between the two (β = −0.047 ml/g/min per mm, 95% confidence interval [CI]: −0.057 to −0.038, P < 0.001) with similar falls in the endocardium and the epicardium (β = −0.048 ml/g/min per mm, 95% CI: −0.058 to −0.039, P < 0.001 and β = −0.048 ml/g/min per mm, 95% CI: −0.058 to −0.038, P < 0.001, respectively).

Segments with LGE were significantly associated with lower perfusion at rest (odds ratio [OR] per ml/g/min increase in MBF: 0.086, 95% CI: 0.078 to 0.095, P = 0.003). This relationship remained consistent at stress (OR: 0.086, 95% CI: 0.081 to 0.092, P < 0.001) and when examined in relation to MPRI (OR: 0.053, 95% CI 0.032 to 0.089, P = 0.015). Both rest and stress MBF appeared to be significantly lower in segments with LGE relative to those without (Figure 5A), resulting in a significant difference in MPRI between segments with and without LGE (MPRI: 1.80 ± 0.74 versus 1.92 ± 0.64, P < 0.001 respectively). However, these differences did not persist after adjusting for differences in wall thickness.

ROI analysis on pixel-level perfusion maps revealed a similar, yet stronger trend of reduced MBF when comparing areas of LGE to remote areas free of LGE. This trend was observed for both rest and stress (Figure 5B). As a result, the MPRI was significantly lower in ROI with LGE than in remote areas (1.61 ± 0.65 versus 1.92 ± 0.54, P < 0.001).

The population studied was drawn from referrals to our clinical service which is a tertiary center, leading to potential selection bias towards higher risk cases. However, patients with implantable cardioverter defibrillators (ICDs) who have been deemed high risk will have been excluded due to the contraindication of CMR in this group potentially counterbalancing this. In addition, 97% of patients had 0 or only 1 risk factor for SCD.

Myocardial fibrosis was assessed using the LGE technique. Whilst this detects replacement fibrosis, it does not allow the quantification of interstitial fibrosis [44]. The association between fibrosis and perfusion abnormalities may therefore have been underappreciated. Nevertheless, replacement fibrosis is thought to be driven by ischemic necrosis and is the distinct type of fibrosis that has been most clearly associated with myocardial ischemia in HCM [4],[5],[24]. Future developments in interstitial imaging using T1-mapping techniques may allow the relationship between interstitial fibrosis, total fibrotic burden and perfusion to be addressed [44].

Finally, we were unable to determine the prognostic significance of our findings given our limited sample size and the relatively low event rate seen in HCM [45]. Nevertheless, our finding of severe microvascular dysfunction in a subgroup of patients with HCM warrants further investigation to determine the potential utility of this phenomenon for risk stratification.